Several approaches to altering the nature of the polyketide product of a PKS by genetic engineering have been proposed: see particularly WO 93/13663 and WO 98/01571. The length of polyketide formed has been altered, in the case of erythromycin biosynthesis, by specific relocation using genetic engineering of the enzymatic domain of the erythromycin-producing PKS that contains the chain-releasing thioesterase/cyclase activity (Cortés, J. et al. Science (1995) 268:1487-1489; Kao, C. M. et al. J. Am. Chem. Soc. (1995) 117:9105-9106).
In-frame deletion of the DNA encoding part of the ketoreductase domain in module 5 of the erythromycin-producing PKS (also known as 6-deoxyerythronolide B synthase, DEBS) has been shown to lead to the formation of erythromycin analogues 5,6-dideoxy-3-α-mycarosyl-5-oxoerythronolide B, 5,6-dideoxy-5-oxoerythronolide B and 5,6-dideoxy, 6 β-epoxy-5-oxoerythronolide B (Donadio, S. et al. Science (1991) 252:675-679). Likewise, alteration of active site residues in the enoylreductase domain of module 4 in DEBS, by genetic engineering of the corresponding PKS-encoding DNA and its introduction into Saccharopolyspora erythraea, led to the production of 6,7-anhydroerythromycin C (Donadio, S. et al. Proc Natl. Acad. Sci. USA (1993) 90:7119-7123).
Patent application WO 00/01827 describes further methods of manipulating a PKS to change the oxidation state of the β-carbon. Substituting the reductive domain of module 2 of the erythromycin-producing PKS with domains derived from rapamycin PKS modules 10 and 13 led to the formation of C10-C11 olefin-erythromycin A and C10-C11 dihydroerythromycin A respectively.
The second class of PKS, named Type II PKSs, is represented by the synthases for aromatic compounds. Type II PKSs contain only a single set of enzymatic activities for chain extension and these are re-used as appropriate in successive cycles (Bibb, M. J. et al. EMBO J. (1989) 8:2727-2736; Sherman, D. H. et al. EMBO J. (1989) 8:2717-2725; Fernandez-Moreno, M. A. et al. J. Biol. Chem. (1992) 267:19278-19290). The “extender” units for the Type II PKSs are usually acetate (malonyl-CoA) units, and the presence of specific cyclases dictates the preferred pathway for cyclisation of the completed chain into an aromatic product (Hutchinson, C. R. and Fujii, I. Annu. Rev. Microbiol. (1995) 49:201-238). Hybrid polyketides have been obtained by the introduction of cloned Type II PKS gene-containing DNA into another strain containing a different Type II PKS gene cluster, for example by introduction of DNA derived from the gene cluster for actinorhodin, a blue-pigmented polyketide from Streptomyces coelicolor, into an anthraquinone polyketide-producing strain of Streptomyces galileus (Bartel, P. L. et al. J. Bacteriol. (1990) 172:4816-4826). Occasionally, unusual starter units are incorporated by Type II PKS, particularly in the biosynthesis of oxytetracycline, frenolicin and daunorubicin and in these cases a separate AT is used to transfer the starter unit to the PKS.
Fungal PKSs such as the 6-methylsalicylic acid or lovastatin PKS typically consist of a single multi-domain polypeptide which include most of the activities required for the synthesis of the polyketide portion of these molecules (Hutchinson C. R. and Fujii I. Annu. Rev. Microbiol. (1995) 49:201-238). Type II Fungal PKSs are also known.
A number of mixed systems comprising polyketide synthase and nonribosomal peptide synthase modules have been identified including the epothilone and bleomycin biosynthetic clusters.
Although large numbers of therapeutically important polyketides have been identified, there remains a need to obtain novel polyketides that have enhanced properties or possess completely novel bioactivity. The complex polyketides produced by Type I PKSs are particularly valuable, in that they include compounds with known utility as anthelminthics, insecticides, anticancer, immunosuppressants, antifungal or antibacterial agents. Because of their structural complexity, such novel polyketides are not readily obtainable by total chemical synthesis, or by chemical modifications of known polyketides. Particular changes that are desired are changes to the carbon skeleton by altering the nature of the starter and/or extender unit(s) incorporated, changes to the oxidation level of the β-keto carbon and therefore the pattern of oxygen substituents by altering the series of reductive steps that occur after chain extension and changes to the post PKS “tailoring” steps which generally comprise hydroxylation, methylation or glycosylation of the polyketide molecule.
There is also a need to develop reliable and specific ways of deploying individual modules in practice so that all, or a large fraction, of hybrid PKS genes that are constructed, are viable and produce the desired polyketide product. Various strategies have been described to produce these hybrid PKSs particularly utilising recombinant DNA technology and denovo biosynthesis. There is a particular need to develop methods of manipulating these PKS in a manner that minimises the alteration to the PKS protein structure. Existing methods of achieving these manipulations sometimes produce hybrid PKS multienzymes which give the desired product at only 1% or less of the rate that the unmodified PKS produces product.
WO 93/13663 and WO 98/01571 describe novel methods of engineering PKSs. A well-established method of altering the nature of the extender unit used at any position in the polyketide molecule, particularly malonyl-, methylmalonyl- or ethylmalonyl-CoA is by domain substitution. For example, WO98/01546 and U.S. Pat. No. 6,063,561 disclose methods of accomplishing this modification to form modified erythromycins. Novel polyketide molecules, in this case particularly novel erythromycins, are produced by the replacement of an entire AT domain-encoding DNA fragment on the Saccharopolyspora erythraea chromosome with an equivalent heterologous AT domain-encoding fragment from another PKS cluster. It is well known to those skilled in the art that selection of the exact DNA/protein splice sites into which to insert the heterologous domain requires detailed analysis of the corresponding DNA and protein sequences. Different researchers choose to use splice sites at conserved, semi-conserved or non-conserved regions of the protein, or at sites either within or at the boundaries of the AT domains. A further drawback of this technique is that it is hard to predict whether a particular heterologous domain will work in any given context. A domain that works successfully in one module may not work at all in an adjoining module or may produce polyketides at a vastly reduced yield. Oliynyk, M. et al. (Chem. Biol. (1996) 3:833-839) and Ruan et al. (J. Bact. (1997) 179:6416-6425) have published studies that exchange a methylmalonyl-CoA specific AT domain for malonyl-CoA specific AT domains in modules of the erythromycin PKS. Products were observed only for changes in modules 1 and 2, with module 2 at a vastly lowered yield. Stassi et al. (Proc. Natl. Acad. Sci. (1998) 95:7305-9) exchange the methylmalonyl-CoA specific AT of module 4 of the erythromycin PKS for an ethylmalonyl-CoA specific AT and again product yield was low even after the addition of the crotonyl-CoA reductase gene thought to increase the supply of the required ethylmalonyl-CoA precursor. A possible reason for the limiting yields is the structural or mechanistic non-compatibility of a heterologous AT domain with the adjoining KS and ACP domains with which it must interact properly for efficient polyketide chain synthesis. Consequently, it is often necessary to try multiple domain swaps to achieve a novel polyketide-producing strain that displays adequate efficiency—a process made particularly arduous when these changes must be made by gene replacement on the chromosome through a two step double integration process. The introduction of splice sites at the DNA level is time consuming and technically challenging, requiring careful analysis to ensure the PKS protein coding reading frame is not disrupted. The introduction of restriction enzyme sites often requires changes at the amino acid level which lead to further PKS protein structure disruption and consequent loss of catalytic efficiency.
A method that could utilise the numerous techniques available for site directed mutagenesis to influence the AT substrate specificity with minimal disruption to the protein tertiary structure would be a valuable addition to the current techniques.
Changes to an active site have been shown to alter substrate specificity in other systems. For example, in an early study, Scrutton et al. (Nature (1990) 343:38-43) used site directed mutagenesis to switch the coenzyme substrate specificity of a glutathione reductase. Identifying and changing a ‘fingerprint’ structural motif in the NADP+ binding domain they could convert the enzyme into one displaying a marked preference for NAD+. The techniques of directed evolution have been used to improve/change enzyme catalytic function. Of many examples in the literature, Zhang et al. (PNAS (1997) 94:4504-4509) illustrate the conversion of a galactosidase to a fucosidase by these techniques. The resulting protein bears 6 mutations, of which 3 lie in, or in close proximity to the active site.
Minor but directed changes to a PKS domain can make significant changes to its catalytic function. Patent application WO 00/00500 teaches that an extender ketosynthase domain is converted to a decarboxylating (and hence loading) ketosynthase domain by site directed mutagenesis at the active site. U.S. Pat. Nos. 6,004,787 and 6,066,721 and Jacobsen et al. Science (1997)277:367-369 describe the deletion of residues in the KS1 active site to inactivate this activity to allow the production of novel polyketides by feeding of synthetic precursors to the modified PKS.
Several studies have attempted to correlate the primary amino acid sequence of the AT to determine amino acids directly involved with the recognition of the appropriate substrate, and particularly the nature of the substrate side chain (i.e. the malonyl portion of the acyl-CoA thioester). Studies by Haydock et al. (FEBS Lett. (1995) 374:246-248) correlated the substrate specificity of malonyl- or methylmalonyl-CoA specific AT with a motif 11 amino acids upstream of the known active site. Comparisons between this motif and the protein structure of a known acyltransferase from E. coli fatty acid synthase allowed the authors to assess the proximity of the motif residues to the active site (and hence its ability to select the substrate). The authors acknowledged that “this divergent region thus identified lies near the acyltransferase active site though not close enough to make direct contact with the substrate”. Other studies (Katz, L. Chem Rev. (1997) 97:2557-2575, Tang, L. et al., Gene (1998) 216:255-265) have correlated additional residues with a specific extender unit using these residues as a tool to predict the AT substrate specificity from a protein sequence derived from polyketide gene cluster sequencing projects. It has remained unclear which residues have mechanistic importance. In only one case have regions within the PKS AT domain been exchanged in an attempt to swap AT specificity; patent application WO 00/01838 and Lau et al. Biochemistry (1999) 38:1643-51) implicated a ‘hypervariable region’ at the C-terminus of the AT domain in the selection of extender unit. These workers interchanged this 25-30 amino acid stretch and showed that this change was sufficient to alter the substrate specificity of the AT, concluding “a short (23-35 amino acid) C-terminal segment present in all AT domains is the principal determinant of their substrate specificity. Interestingly its length and amino acid sequence vary considerably among the known AT domains. We therefore suggest that the choice of extender units by the PKS modules is influenced by a “hypervariable region”, which could be manipulated via combinatorial mutagenesis to generate novel AT domains possessing relaxed or altered substrate specificity”. Surprisingly, our structure molecular modelling studies indicate this region lies at a surface accessible region away from the active site and hence is unlikely to directly interact with (and hence directly select) the malonyl portion or the substrate used. The effect on substrate specificity is therefore likely to be imprecise and due to more indirect effects via, for example, disruption of tertiary structure.